Test Accuracy Systematic Review Research Articles

Use of artificial intelligence for image analysis in breast cancer screening programmes: systematic review of test accuracy

ObjectiveTo examine the accuracy of artificial intelligence (AI) for the detection of breast cancer in mammography screening practice.DesignSystematic review of test accuracy studies.Data sourcesMedline, Embase, Web of Science, and Cochrane...

Polyarteritis Nodosa: A Systematic Review of Test Accuracy and Benefits and Harms of Common Treatments.

ObjectiveThe object of this study was to analyze the benefits and harms of different treatment options and to analyze test accuracy used in the evaluation of patients with primary systemic polyarteritis nodosa (PAN).MethodsA systematic search of published English‐language literature was performed in Ovid Medline, PubMed, Embase, and the Cochrane Library from the inception of each database through August 2019. Articles were screened for suitability in addressing patient, intervention, comparison, and outcome questions, with studies presenting the highest level of evidence given preference.ResultsOf 137 articles selected for data abstraction, we analyzed 21 observational studies and seven randomized controlled trials (RCTs). The results showed indirect evidence that a deep skin biopsy provides good diagnostic accuracy. A combined nerve and muscle biopsy should be obtained for patients with PAN with peripheral neuropathy. Cyclophosphamide with high‐dose glucocorticoids (GCs) is effective as an induction treatment for newly diagnosed active and severe PAN. GC monotherapy is adequate in the majority of patients with nonsevere PAN, although it has a high relapse rate with GC taper. There was insufficient data in determining the optimal duration of non‐GC and GC maintenance therapy. Tumor necrosis factor inhibitors are effective treatment for patients with deficiency of adenosine deaminase 2 (DADA2) with stroke and vasculitis manifestations.ConclusionThis comprehensive systematic review synthesizes and evaluates the harms and benefits of different treatment options and the accuracy of commonly used tests for the diagnosis of systemic PAN. Data for diagnosis and management of PAN and DADA2 are mostly limited to observational studies. More high‐quality RCTs are needed.

Informant-based screening tools for diagnosis of dementia, an overview of systematic reviews of test accuracy studies protocol

BackgroundRobust diagnosis of dementia requires an understanding of the accuracy of the available diagnostic tests. Informant questionnaires are frequently used to assess for dementia in clinical practice. Recent systematic reviews have sought to establish the diagnostic test accuracy of various dementia informant screening tools. However, most reviews to date have focused on a single diagnostic tool and this does not address which tool is ‘best’. A key aim of the overview of systematic reviews is to present a disparate evidence base in a single, easy to access platform.MethodsWe will conduct an overview of systematic reviews in which we ‘review the systematic reviews’ of diagnostic test accuracy studies evaluating informant questionnaires for dementia. As an overview of systematic reviews of test accuracy is a relatively novel approach, we will use this review to explore methods for visual representation of complex data, for highlighting evidence gaps and for indirect comparative analyses. We will create a list of informant tools by consulting with dementia experts. We will search 6 databases (EMBASE (OVID); Health and Psychosocial Instruments (OVID); Medline (OVID); CINAHL (EBSCO); PSYCHinfo (EBSCO) and the PROSPERO registry of review protocols) to identify systematic reviews that describe the diagnostic test accuracy of informant questionnaires for dementia. We will assess review quality using the AMSTAR-2 (Assessment of Multiple Systematic Reviews) and assess reporting quality using PRISMA-DTA (Preferred Reporting Items for Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies) checklists. We will collate the identified reviews to create an ‘evidence map’ that highlights where evidence does and does not exist in relation to informant questionnaires. We will pool sensitivity and specificity data via meta-analysis to generate a diagnostic test accuracy summary statistic for each informant questionnaire. If data allow, we will perform a statistical comparison of the diagnostic test accuracy of each informant questionnaire using a network approach.DiscussionOur overview of systematic reviews will provide a concise summary of the diagnostic test accuracy of informant tools and highlight areas where evidence is currently lacking in this regard. It will also apply network meta-analysis techniques to a new area.

Detecting delirium in patients with acute stroke: a systematic review of test accuracy

BackgroundPatients with acute stroke are particularly vulnerable to delirium episodes. Although delirium detection is important, no evidence-based recommendations have been established to date on how these patients should be routinely screened for delirium or which tool should be used for this purpose in this population. Therefore, the aim of this study was to identify delirium screening tools for patients with acute stroke and to summarise their accuracy.MethodsFollowing the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, a systematic search of Medline, CINAHL and Scopus databases was performed to include: (a) diagnostic test accuracy studies; (b) evaluating tools detecting delirium among patients with acute stroke; (c) written in English; (d) published up to September 2018. The included studies were assessed in their quality by using the Quality Assessment of Diagnostic Accuracy Studies-2.ResultsA total of four studies have been performed to date in the field with a variable quality for the methodology used and documentation of the accuracy of mainly two tools, as (1) the 4-Assessment Test for delirium (4AT), reporting a range of sensitivity from 90.2 to 100% and a specificity from 64.5 to 86%; and (2) the Confusion Assessment Method-Intensive Care Unit (CAM-ICU) showing a sensitivity of 76% (95% Confidence of Interval [CI] 55–91) and a specificity of 98% (95%CI 93–100). Other tools have been studied as: The Abbreviated Mental Test-10, the Abbreviated Mental Test short form, the Clock Drawing Test, the Cognitive Examination derived from the National Institutes of Health Stroke Scale and the Glasgow Coma Scale. Moreover, the use of a single question—namely, ‘Does this patient have cognitive issues?’ as answered by the multidisciplinary team—has been subjected to a validation process.ConclusionsTo date a few primary studies have been published to test the accuracy of tools in their ability to detect post-stroke delirium; among those available, the 4AT and the CAM-ICU tools have been mostly studied. Research has just started to add evidence to the challenge of detecting and usefully assessing newly-acquired delirium among stroke patients: therefore, more studies are needed to improve the knowledge and allow a robust selection of the most useful tool to use in this population.

18F-FDG PET for Prediction of Conversion to Alzheimer's Disease Dementia in People with Mild Cognitive Impairment: An Updated Systematic Review of Test Accuracy.

Background:A previous Cochrane systematic review concluded there is insufficient evidence to support the routine use of 18F-FDG PET in clinical practice in people with mild cognitive impairment (MCI).Objectives:To update the evidence and reassess the accuracy of 18F-FDG-PET for detecting people with MCI at baseline who would clinically convert to Alzheimer’s disease (AD) dementia at follow-up.Methods:A systematic review including comprehensive search of electronic databases from January 2013 to July 2017, to update original searches (1999 to 2013). All key review steps, including quality assessment using QUADAS 2, were performed independently and blindly by two review authors. Meta-analysis could not be conducted due to heterogeneity across studies.Results:When all included studies were examined across all semi-quantitative and quantitative metrics, exploratory analysis for conversion of MCI to AD dementia (n = 24) showed highly variable accuracy; half the studies failed to meet four or more of the seven sets of QUADAS 2 criteria. Variable accuracy for all metrics was also found across eleven newly included studies published in the last 5 years (range: sensitivity 56–100%, specificity 24–100%). The most consistently high sensitivity and specificity values (approximately ≥80%) were reported for the sc-SPM (single case statistical parametric mapping) metric in 6 out of 8 studies.Conclusion:Systematic and comprehensive assessment of studies of 18FDG-PET for prediction of conversion from MCI to AD dementia reveals many studies have methodological limitations according to Cochrane diagnostic test accuracy gold standards, and shows accuracy remains highly variable, including in the most recent studies. There is some evidence, however, of higher and more consistent accuracy in studies using computer aided metrics, such as sc-SPM, in specialized clinical settings. Robust, methodologically sound prospective longitudinal cohort studies with long (≥5 years) follow-up, larger consecutive samples, and defined baseline threshold(s) are needed to test these promising results. Further evidence of the clinical validity and utility of 18F-FDG PET in people with MCI is needed.

Fundus autofluorescence imaging: systematic review of test accuracy for the diagnosis and monitoring of retinal conditions.

We conducted a systematic review of the accuracy of fundus autofluorescence (FAF) imaging for diagnosing and monitoring retinal conditions. Searches in November 2014 identified English language references. Sources included MEDLINE, EMBASE, the Cochrane Library, Web of Science, and MEDION databases; reference lists of retrieved studies; and internet pages of relevant organisations, meetings, and trial registries. For inclusion, studies had to report FAF imaging accuracy quantitatively. Studies were critically appraised using QUADAS risk of bias criteria. Two reviewers conducted all review steps. From 2240 unique references identified, eight primary research studies met the inclusion criteria. These investigated diagnostic accuracy of FAF imaging for choroidal neovascularisation (one study), reticular pseudodrusen (three studies), cystoid macular oedema (two studies), and diabetic macular oedema (two studies). Diagnostic sensitivity of FAF imaging ranged from 32 to 100% and specificity from 34 to 100%. However, owing to methodological limitations, including high and/or unclear risks of bias, none of these studies provides conclusive evidence of the diagnostic accuracy of FAF imaging. Study heterogeneity precluded meta-analysis. In most studies, the patient spectrum was not reflective of those who would present in clinical practice and no studies adequately reported whether FAF images were interpreted consistently. No studies of monitoring accuracy were identified. An update in October 2016, based on MEDLINE and internet searches, identified four new studies but did not alter our conclusions. Robust quantitative evidence on the accuracy of FAF imaging and how FAF images are interpreted is lacking. We provide recommendations to address this.

OC-7a: Imaging for the exclusion of pulmonary embolism in pregnancy – a Cochrane Systematic Review of diagnostic test accuracy

OC-7a: Imaging for the exclusion of pulmonary embolism in pregnancy – a Cochrane Systematic Review of diagnostic test accuracy

Defining the clinical pathway in cochrane diagnostic test accuracy reviews.

BackgroundThe value of a medical test depends on the context in which it might be used. Ideally, questions, results and conclusions of a diagnostic test accuracy (DTA) systematic review should be presented in light of this context. There is increasing acceptance of the value for knowing the impact a test can have on downstream consequences such as costs, implications for further testing and treatment options however there is currently no explicit guidance on how to address this. Authors of a Cochrane diagnostic review have recently been asked to include the clinical pathway in which a test maybe used. We aimed to evaluate how authors were developing their clinical pathways in the light of this.MethodsWe searched the Cochrane Database of Systematic Reviews for all published DTA reviews. We included only those reviews that included a clinical pathway. We developed a checklist, based on the guidance in the Cochrane Handbook for DTA review authors. To this, we added a number of additional descriptors. We checked if the included pathways fulfilled these descriptors as defined by our checklist.ResultsWe found 47 reviews, of which 33 (73 %) contained aspects pertaining to a clinical pathway. The 33 reviews addressed the clinical pathway differently, both in content and format. Of these, 21 provided a textual description and 12 include visual and textual descriptions. There was considerable variation in how comprehensively review authors adhered to our checklist. Eighteen reviews (51 %) linked the index test results to downstream clinical management actions and patient consequences, but only eight went on to differentially report on the consequences for false negative results and nine on the consequences for false positive results.ConclusionThere is substantial variation in the clinical pathway descriptions in Cochrane systematic reviews of test accuracy. Most reviews do not link misclassifications (i.e. false negatives and false positive) to downstream patient consequences. Review authors could benefit from more explicit guidance on how to create such pathways, which in turn can help guide them in their evidence selection and appraisal of the evidence in the context of downstream consequences of testing.Electronic supplementary materialThe online version of this article (doi:10.1186/s12874-016-0252-x) contains supplementary material, which is available to authorized users.

Placental growth factor (alone or in combination with soluble fms-like tyrosine kinase 1) as an aid to the assessment of women with suspected pre-eclampsia: systematic review and economic analysis.

Pre-eclampsia (PE) prediction based on blood pressure, presence of protein in the urine, symptoms and laboratory test abnormalities can result in false-positive diagnoses. This may lead to unnecessary antenatal admissions and preterm delivery. Blood tests that measure placental growth factor (PlGF) or the ratio of soluble fms-like tyrosine kinase 1 (sFlt-1) to PlGF could aid prediction of PE if either were added to routine clinical assessment or used as a replacement for proteinuria testing. To evaluate the diagnostic accuracy and cost-effectiveness of PlGF-based tests for patients referred to secondary care with suspected PE in weeks 20-37 of pregnancy. Systematic reviews and an economic analysis. Bibliographic databases including MEDLINE, EMBASE, Web of Science and The Cochrane Library and Database of Abstracts of Reviews of Effects were searched up to July 2015 for English-language references. Conferences, websites, systematic reviews and confidential company submissions were also accessed. Systematic reviews of test accuracy and economic studies were conducted to inform an economic analysis. Test accuracy studies were required to include women with suspected PE and report quantitatively the accuracy of PlGF-based tests; their risk of bias was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) criteria. The economic studies review had broad eligibility criteria to capture any types of economic analysis; critical appraisal employed standard checklists consistent with National Institute for Health and Care Excellence criteria. Study selection, critical appraisal and data extraction in both reviews were performed by two reviewers. An independent economic analysis was conducted based on a decision tree model, using the best evidence available. The model evaluates costs (2014, GBP) from a NHS and Personal Social Services perspective. Given the short analysis time horizon, no discounting was undertaken. Four studies were included in the systematic review of test accuracy: two on Alere's Triage® PlGF test (Alere, Inc., San Diego, CA, USA) for predicting PE requiring delivery within a specified time and two on Roche Diagnostics' Elecsys® sFlt-1 to PlGF ratio test (Roche Diagnostics GmbH, Mannheim, Germany) for predicting PE within a specified time. Three studies were included in the systematic review of economic studies, and two confidential company economic analyses were assessed separately. Study heterogeneity precluded meta-analyses of test accuracy or cost-analysis outcomes, so narrative syntheses were conducted to inform the independent economic model. The model predicts that, when supplementing routine clinical assessment for rule-out and rule-in of PE, the two tests would be cost-saving in weeks 20-35 of gestation, and marginally cost-saving in weeks 35-37, but with minuscule impact on quality of life. Length of neonatal intensive care unit stay was the most influential parameter in sensitivity analyses. All other sensitivity analyses had negligible effects on results. No head-to-head comparisons of the tests were identified. No studies investigated accuracy of PlGF-based tests when used as a replacement for proteinuria testing. Test accuracy studies were found to be at high risk of clinical review bias. The Triage and Elecsys tests would save money if added to routine clinical assessment for PE. The magnitude of savings is uncertain, but the tests remain cost-saving under worst-case assumptions. Further research is required to clarify how the test results would be interpreted and applied in clinical practice. This study is registered as PROSPERO CRD42015017670. The National Institute for Health Research Health Technology Assessment programme.

Diagnosis of acute rhinosinusitis in primary care: a systematic review of test accuracy.

Acute rhinosinusitis (ARS) is a common primary care infection, but there have been no recent, comprehensive diagnostic meta-analyses. To determine the accuracy of laboratory and imaging studies for the diagnosis of ARS. Systematic review of diagnostic tests in outpatient, primary care, and specialty settings. The authors included studies of patients presenting with or referred for suspected ARS, and used bivariate meta-analysis to calculate summary estimates of test accuracy and the area under the receiver operating characteristic (ROC) curve. The authors also plotted summary ROC curves to explore heterogeneity, cutoffs, and the impact of different reference standards. Using antral puncture as the reference standard, A mode ultrasound (positive likelihood ratio [LR+] 1.71, negative likelihood ratio [LR-] 0.41), B mode ultrasound (LR+ 1.64, LR- 0.69), and radiography (LR+ 2.01, LR- 0.28) had only modest accuracy. Accuracy was higher using imaging as the reference standard for both ultrasound (LR+12.4, LR- 0.35) and radiography (LR+ 9.4, LR- 0.27), although this likely overestimates accuracy. C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR) both had clear threshold effects, and modest overall accuracy. The LR+ for ESR >30 and >40 were 4.08 and 7.40, respectively. A dipstick of nasal secretions for leucocyte esterase was highly accurate (LR+ 18.4, LR- 0.17) but has not been validated. In general, tests were of limited value in the diagnosis of ARS. Normal radiography helps rule out sinusitis when negative, whereas CRP and ESR help rule in sinusitis when positive, although, given their limited accuracy as individual tests, they cannot be routinely recommended. Prospective studies integrating signs and symptoms with point-of-care CRP, dipstick, and/ or handheld B-mode ultrasound are needed.

Decision-Making about Healthcare Related Tests and Diagnostic Strategies: User Testing of GRADE Evidence Tables.

ObjectiveTo develop guidance on what information to include and how to present it in tables summarizing the evidence from systematic reviews of test accuracy following the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach.MethodsTo design and refine the evidence tables, we used an iterative process based on the analysis of data from four rounds of discussions, feedback and user testing. During the final round, we conducted one-on-one user testing with target end users. We presented a number of alternative formats of evidence tables to participants and obtained information about users’ understanding and preferences.ResultsMore than 150 users participated in initial discussions and provided their formal and informal feedback. 20 users completed one-on-one user testing interviews. Almost all participants preferred summarizing the results of systematic reviews of test accuracy in tabular format rather than plain text. Users generally preferred less complex tables but found presenting sensitivity and specificity estimates only as too simplistic. Users found the presentation of test accuracy for several values of prevalence initially confusing but modifying table layout and adding sample clinical scenarios for each prevalence reduced this confusion. Providing information about clinical consequences of testing result was viewed as not feasible for authors of systematic reviews.ConclusionWe present the current formats for tables presenting test accuracy following the GRADE approach. These tables can be developed using GRADEpro guidelines development tool (www.guidelinedevelopment.org or www.gradepro.org) and are being further developed into electronic interactive tables that will suit the needs of different end users. The formatting of these tables, and how they influence result interpretation and decision-making will be further evaluated in a randomized trial.

Incorporating quality assessments of primary studies in the conclusions of diagnostic accuracy reviews: a cross-sectional study

BackgroundDrawing conclusions from systematic reviews of test accuracy studies without considering the methodological quality (risk of bias) of included studies may lead to unwarranted optimism about the value of the test(s) under study. We sought to identify to what extent the results of quality assessment of included studies are incorporated in the conclusions of diagnostic accuracy reviews.MethodsWe searched MEDLINE and EMBASE for test accuracy reviews published between May and September 2012. We examined the abstracts and main texts of these reviews to see whether and how the results of quality assessment were linked to the accuracy estimates when drawing conclusions.ResultsWe included 65 reviews of which 53 contained a meta-analysis. Sixty articles (92%) had formally assessed the methodological quality of included studies, most often using the original QUADAS tool (n = 44, 68%). Quality assessment was mentioned in 28 abstracts (43%); with a majority (n = 21) mentioning it in the methods section. In only 5 abstracts (8%) were results of quality assessment incorporated in the conclusions. Thirteen reviews (20%) presented results of quality assessment in the main text only, without further discussion. Forty-seven reviews (72%) discussed results of quality assessment; the most frequent form was as limitations in assessing quality (n = 28). Only 6 reviews (9%) further linked the results of quality assessment to their conclusions, 3 of which did not conduct a meta-analysis due to limitations in the quality of included studies. In the reviews with a meta-analysis, 19 (36%) incorporated quality in the analysis. Eight reported significant effects of quality on the pooled estimates; in none of them these effects were factored in the conclusions.ConclusionWhile almost all recent diagnostic accuracy reviews evaluate the quality of included studies, very few consider results of quality assessment when drawing conclusions. The practice of reporting systematic reviews of test accuracy should improve if readers not only want to be informed about the limitations in the available evidence, but also on the associated implications for the performance of the evaluated tests.

Systematic review of the use of bone turnover markers for monitoring the response to osteoporosis treatment: the secondary prevention of fractures, and primary prevention of fractures in high-risk groups

There is currently no standard practice for the monitoring of patients receiving treatment for osteoporosis. Repeated dual-energy X-ray absorptiometry (DXA) is commonly used for monitoring treatment response, but it has its limitations. Bone turnover markers have advantages over DXA as they are non-invasive, relatively cheap and can detect changes in bone turnover rates earlier. However, they do have disadvantages, particularly high within- and between-patient variability. The ability of bone turnover markers to identify treatment non-responders and predict future fracture risk has yet to be established. We aimed to determine the clinical effectiveness, test accuracy, reliability, reproducibility and cost-effectiveness of bone turnover markers for monitoring the response to osteoporosis treatment. We searched 12 electronic databases (including MEDLINE, EMBASE, The Cochrane Library and trials registries) without language restrictions from inception to March 2012. We hand-searched three relevant journals for the 12 months prior to May 2012, and websites of five test manufacturers and the US Food and Drug Administration (FDA). Reference lists of included studies and relevant reviews were also searched. A systematic review of test accuracy, clinical utility, reliability and reproducibility, and cost-effectiveness of two formation and two resorption bone turnover markers, in patients being treated for osteoporosis with any of bisphosphonate [alendronate (Fosamax, MSD), risedronate (Actonel, Warner Chilcott Company), zolendronate (Zometa, Novartis)], raloxifene (Evista, Eli Lilly and Company Ltd), strontium ranelate (Protelos, Servier Laboratories Ltd), denosumab (Prolia, Amgen Ltd) or teriparatide (Forsteo, Eli Lilly and Company Ltd), was undertaken according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines. Given the breadth of the review question, a range of study designs and outcome measures were eligible. The development of a decision model was planned to determine the cost-effectiveness of bone turnover markers for informing changes in patient management if clinical effectiveness could be established. Forty-two studies (70 publications) met the inclusion criteria; none evaluated cost-effectiveness. Only five were randomised controlled trials (RCTs); these assessed only the impact of bone marker monitoring on aspects of adherence. No RCTs evaluated the effectiveness of bone turnover marker monitoring on treatment management. One trial suggested that feedback of a good response decreased non-persistence [hazard ratio (HR) 0.71, 95% confidence interval (CI) 0.53 to 0.95], and feedback of a poor response increased non-persistence (HR 2.22, 95% CI 1.27 to 3.89); it is not clear whether or not the trial recruited a population representative of that seen in clinical practice. Thirty-three studies reported results of some assessment of test accuracy, mostly correlations between changes in bone turnover and bone mineral density. Only four studies reported on intra- or interpatient reliability and reproducibility in treated patients. Overall, the results were inconsistent and inconclusive, owing to considerable clinical heterogeneity across the studies and the generally small sample sizes. As clinical effectiveness of bone turnover monitoring could not be established, a decision-analytic model was not developed. There was insufficient evidence to inform the choice of which bone turnover marker to use in routine clinical practice to monitor osteoporosis treatment response. The research priority is to identify the most promising treatment-test combinations for evaluation in subsequent, methodologically sound, RCTs. In order to determine whether or not bone turnover marker monitoring improves treatment management decisions, and ultimately impacts on patient outcomes in terms of reduced incidence of fracture, RCTs are required. Given the large number of potential patient population-treatment-test combinations, the most promising combinations would initially need to be identified in order to ensure that any RCTs focus on evaluating those strategies. As a result, the research priority is to identify these promising combinations, by either conducting small variability studies or initiating a patient registry to collect standardised data. The National Institute for Health Research Health Technology Assessment programme.

Sentinel lymph node status in vulval cancer: systematic reviews of test accuracy and decision-analytic model-based economic evaluation

Vulval cancer causes 3-5% of all gynaecological malignancies and requires surgical removal and inguinofemoral lymphadenectomy (IFL). Complications affect > 50% of patients, including groin wound infection, lymphoedema and cellulitis. A sentinel lymph node (SLN) is the first groin node with the highest probability of malignancy. SLN biopsy would be useful if it could accurately identify patients in whom cancer has spread to the groin, without removing all groin nodes. SLNs can be identified by isosulfan blue dye and/or technetium-99 ((99m)Tc) radioactive tracer during lymphoscintigraphy. The blue dye/(99m)Tc procedure only detects SLN, not metastases - this requires histological examination, which can include ultrastaging and staining with conventional haematoxylin and eosin (H&E) or immunohistochemistry. To determine the test accuracy and cost-effectiveness of the SLN biopsy with (99m)Tc and/or blue dye compared with IFL or clinical follow-up for test negatives in vulval cancer, through systematic reviews and economic evaluation. Standard medical databases, including MEDLINE, EMBASE, Science Citation Index and The Cochrane Library, medical search gateways, reference lists of review articles and included studies were searched to January 2011. For accuracy and effectiveness, standard methods were used and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Searches were to January 2011, with no language restrictions. Meta-analyses were carried out with Meta-Disc version 1.4 (Javier Zamora, Madrid, Spain) for accuracy; none was appropriate for effectiveness. The economic evaluation from a NHS perspective used a decision-tree model in DATA TreeAge Pro Healthcare 2001 (TreeAge Software, Inc., Williamstown, MA, USA). Six options (blue dye with H&E, blue dye with ultrastaging, (99m)Tc with H&E, (99m)Tc with ultrastaging, blue dye/(99m)Tc with H&E, blue dye/(99m)Tc with ultrastaging) were compared with IFL. Deterministic and probabilistic sensitivity analyses were conducted. For accuracy, of the 26 included studies, most evaluated (99m)Tc/blue dye combined. Four studies had clinical follow-up only for test negatives and five had clinical follow-up for all and IFL for test negatives. Numbers with no SLN found were difficult to distinguish from those with negative SLN biopsies. The largest group of 11 studies using (99m)Tc/blue dye, ultrastaging and immunohistochemistry had a pooled sensitivity of 95.6% [95% confidence interval (CI) 91.5% to 98.1%] and a specificity of 100% (95% CI 99.0% to 100%). Mean SLN detection rates were 94.6% for (99m)Tc, 68.7% for blue dye and 97.7% for both. One study measured global health status quality of life (QoL) and found no difference between SLN biopsy and IFL. One patient preference evaluation showed that 66% preferred IFL rather than a 5% false-negative rate from SLN biopsy. For effectiveness, of 14,038 references, one randomised controlled trial, three case-control studies and 13 case series were found. Approximately 50% died from vulval cancer and 50% from other causes during follow-ups. Recurrences were in the ratio of approximately 4 : 2 : 1 vulval, groin and distant, with more recurrences in node-positive patients. No studies reported QoL. For cost per death averted, IFL was less costly and more effective than strategies using SLN biopsy. For morbidity-free survival and long-term morbidity-free survival, (99m)Tc with ultrastaging was most cost-effective. Strategies with blue dye only and H&E only were never cost-effective. The incremental cost-effectiveness ratio for (99m)Tc with ultrastaging compared with IFL was £4300 per case of morbidity-free survival and £7100 per long-term morbidity-free survival. The main limitations of this study include the lack of good-quality evidence on accuracy, effectiveness and QoL. A large project such as this takes time to publish, so the most recent studies are not included. A sensitive and specific combined metastatic SLN detection test and information on generic QoL in vulval cancer is urgently required. The National Institute for Health Research Health Technology Assessment programme.

Rapid fetal fibronectin testing to predict preterm birth in women with symptoms of premature labour: a systematic review and cost analysis

Premature birth is defined as birth of before 37 completed weeks' gestation. Not all pregnant women showing symptoms of preterm labour will go on to deliver before 37 weeks' gestation. Hence, addition of fetal fibronectin (fFN) testing to the diagnostic workup of women with suspected preterm labour may help to identify those women who do not require active management, and thus avoid unnecessary interventions, hospitalisations and associated costs. To assess the clinical effectiveness and cost-effectiveness of rapid fFN testing in predicting preterm birth (PTB) in symptomatic women. Bibliographic databases (including EMBASE, Cochrane Database of Systematic Reviews and Cochrane Central Register of Controlled Trials) were searched from 2000 to September/November 2011. Trial registers were also searched. Systematic review methods followed published guidance; we assessed clinical effectiveness and updated a previous systematic review of test accuracy. Risk of bias was assessed using the Cochrane tool (randomised controlled trials; RCTs) and a modification of QUADAS-2 (diagnostic test accuracy studies; DTAs). Summary risk ratios or weighted mean difference were calculated using random-effects models. Summary sensitivity and specificity used a bivariate summary receiver operating characteristic model. Heterogeneity was investigated using subgroup and sensitivity analyses. Health economic analysis focused on cost consequences. The time horizon was hospital admission for observation. A main structural assumption was that, compared with usual care, fFN testing doesn't increase adverse events or negative pregnancy outcomes. Five RCTs and 15 new DTAs were identified. No RCT reported significant effects of fFN testing on maternal or neonatal outcomes. One study reported a subgroup analysis of women with negative fFN test observed > 6 hours, which showed a reduction in length of hospital stay where results were known to clinicians. Combining data from new studies and the previous systematic review, the pooled estimates of sensitivity and specificity were: 76.7% and 82.7% for delivery within 7-10 days of testing; 69.1% and 84.4% for delivery < 34 weeks' gestation; and 60.8% and 82.3% for delivery < 37 weeks' gestation. Estimates were similar across all subgroups sensitivity analyses. The base-case cost analysis resulted in a cost saving of £23.87 for fFN testing compared with usual care. The fFN testing was cost-neutral at an approximate cost of £45. Probabilistic sensitivity analysis gave an incremental cost (saving) of -£25.59 (97.5% confidence interval -£304.96 to £240.06), indicating substantial uncertainty. Sensitivity analyses indicated that admission rate had the largest impact on results. Fetal fibronectin testing has moderate accuracy for predicting PTB. The main potential role is likely to be reducing health-care resource usage by identifying women not requiring intervention. Evidence from RCTs suggests that fFN does not increase adverse outcomes and may reduce resource use. The base-case analysis showed a modest cost difference in favour of fFN testing, which is largely dependent on whether or not fFN testing reduces hospital admission. Currently, there are no high-quality studies and the existing trials were generally underpowered. Hence, there is a need for high-quality adequately powered trials using appropriate study designs to confirm the findings presented. PROSPERO 2011:CRD42011001468. Available from www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42011001468. The National Institute for Health Research Health Technology Assessment programme.

Surveillance After Treatment for Cervical Intraepithelial Neoplasia

To estimate outcomes and costs of surveillance strategies after treatment for high-grade cervical intraepithelial neoplasia (CIN). A hypothetical cohort of women was evaluated after treatment for CIN 2 or 3 using a Markov model incorporating data from a large study of women treated for CIN, systematic reviews of test accuracy, and individual preferences. Surveillance strategies included initial conventional or liquid-based cytology, human papillomavirus testing, or colposcopy 6 months after treatment, followed by annual or triennial cytology. Estimated outcomes included CIN, cervical cancer, cervical cancer deaths, life expectancy, costs, cost per life-year, and cost per quality-adjusted life-year. Conventional cytology at 6 and 12 months, followed by triennial cytology, was least costly. Compared with triennial cytology, annual cytology follow-up reduced expected cervical cancer deaths by 73% to 77% and had an average incremental cost per life-year gained of $69,000 to $81,000. For colposcopy followed by annual cytology, the incremental cost per life-year gained ranged from $70,000 to more than $1 million, depending on risk. Between-strategy differences in mean additional life expectancy per woman were less than 4 days; differences in mean incremental costs per woman were as high as $822. In the cost-utility analysis, colposcopy at 6 months followed by annual cytology had an incremental cost per quality-adjusted life-year of less than $5,500. Human papillomavirus testing or liquid-based cytology added little to no improvement to life-expectancy with higher costs. Annual conventional cytology surveillance reduced cervical cancers and cancer deaths compared with triennial cytology. For high risk of recurrence, a strategy of colposcopy at 6 months increased life expectancy and quality-adjusted life expectancy. Human papillomavirus testing and liquid-based cytology increased costs, but not effectiveness, compared with traditional approaches.

Antenatal ultrasound to predict postnatal renal function in congenital lower urinary tract obstruction: systematic review of test accuracy

Congenital urinary tract obstruction can lead to perinatal mortality and morbidity. Assessing antenatal ultrasound features may help identify fetuses likely to suffer complications. To evaluate the accuracy of antenatal ultrasound in the prediction of postnatal renal function in cases of congenital lower urinary tract obstruction (LUTO). We conducted electronic searches in MEDLINE, EMBASE, the Cochrane Library, MEDION, SIGLE, SCISEARCH Index of scientific and technical proceedings, National Research Register (database inception - 2008) and Medical Conferences register and searched reference lists. Two reviewers independently selected articles in which the accuracy of fetal ultrasound features were evaluated to predict postnatal renal function with no language restrictions. Data were extracted on study characteristics, quality and results to construct 2 x 2 tables. Likelihood ratios for positive (LR+) and negative (LR-) test results, sensitivity and specificity were generated for the different ultrasound parameters and reference standards. Thirteen articles that met the selection criteria, including 215 women and 33 2 x 2 tables. Meta-analysis was performed using clinically similar subgroups to minimise clinical heterogeneity. The ultrasound parameter that showed the best predictive value for postnatal renal function in survivors was renal cortical appearance, sensitivity 0.57 (95% CI 0.37-0.76) and specificity 0.84 (95% CI 0.71-0.94), area under the curve 0.78. Measurement of amniotic fluid volume and the appearance of the renal cortex at diagnosis of LUTO show promising predictive accuracy for poor postnatal renal function.

Gynaecologists blaze the trail in primary studies and systematic reviews of diagnostic test accuracy

As the Cochrane Collaboration is poised to begin publishing systematic reviews of diagnostic test accuracy studies in addition to its traditional systematic reviews of treatment effectiveness, we are likely to see a major expansion in the number of primary studies and systematic reviews of diagnostic test accuracy in the medical literature. Obstetricians and gynaecologists have played an important role in initiating this newer area of research. However, the methodology for such studies is challenging and the published literature is riddled with pitfalls. This editorial seeks to simplify the concepts involved in diagnostic test accuracy studies and systematic reviews, to reflect on the early development of this research in our specialty and to envision the future pathway for screening and diagnostic research.

Locating systematic reviews of test accuracy studies: How five specialist review databases measure up

The aim of this study was to examine location of systematic reviews of test accuracy in five specialist review databases: York CRD's DARE and HTA databases, Medion (University of Maastricht), C-EBLM (International Federation of Clinical Chemistry), and the ARIF in-house database (University of Birmingham). Searches were limited to the period 1996-2006. Test accuracy reviews were located using in-house diagnostic search filters and with help from database producers where databases were not confined to test accuracy reviews. References were coded according to disease area, review purpose, and test application. Ease of use, volume, overlap, and content of databases was noted. A large degree of overlap existed between databases. Medion contained the largest number (n = 672) and the largest number of unique (n = 328) test accuracy references. A combination of three databases identified only 76% of test reviews. All databases were rated as easy to search but varied with respect to timeliness and compatibility with reference management software. Most reviews evaluated test accuracy (85%) but the HTA database had a larger proportion of cost-effectiveness and screening reviews and C-EBLM more reviews addressing early test development. Most reviews were conducted in secondary care settings. Specialist review databases offer an essential addition to general bibliographic databases where application of diagnostic method filters can compromise search sensitivity. Important differences exist between databases in terms of ease of use and content. Our findings raise the question whether the current balance of research setting, in particular the predominance of research on tests used in secondary care, matches the needs of decision makers.

Systematic reviews of test accuracy should search a range of databases to identify primary studies

Objective To estimate the yield from searching a range of bibliographic databases and additional sources to identify test accuracy studies for systematic reviews. Study Design and Setting We examined eight systematic reviews and their database searches: MEDLINE, EMBASE, BIOSIS, Science Citation Index, LILACS, Pascal, and CENTRAL. We used studies included in each systematic review as the “gold standard,” against which yield was estimated. For each database, we classified studies in each gold standard set as being (1) included in the database and identified by searches, (2) included and not identified, and (3) not included in the database. Results No search identified all studies in any gold standard set. EMBASE, Science Citation Index, and BIOSIS contained studies that were not on MEDLINE. Over 20% of studies in the gold standard sets were not identified by searching MEDLINE. Six studies on LILACS were not on any other database. Eight gold standard studies were not included in any of the databases, and a further 22 were not identified by the electronic search strategies. Conclusions Systematic reviews of test accuracy studies should search a range of databases. Even searches designed to be very sensitive, that do not use study design filters, can fail to identify relevant studies.